Video to MP3 (Extract Audio)

Continue in the tool

More pages for this tool

boost the volume of a video's audio mp3

Boost the volume of a video's audio while extracting it as MP3

Under More options, turn on Normalize the volume. The tool measures the section, shows the current loudness, the peak and the gain in dB, and applies a single gain while extracting.

Perceived loudness, peak and what "normalize" changes

Two numbers describe audio. The peak, in dBFS, is the highest sample: 0 dBFS is the digital maximum and going past it is distortion. Perceived loudness, in LUFS (ITU-R BS.1770), measures energy over time with a curve that mimics the ear and ignores silences; that is what tells you whether a video sounds quiet. A video can peak at -3 dBFS, looking "at the limit", and still sit at -24 LUFS, far below streaming's -14 LUFS reference. Raising volume without looking at the peak clips; looking only at the peak leaves the video quiet. LUFS normalization handles both.
The tool applies a constant gain: target minus measured loudness. The targets are -14 LUFS (streaming reference), -16 LUFS (podcast) and -23 LUFS (TV, EBU R128). The gain never takes the peak above -1 dBFS: if the target needs more, it stops at the ceiling and the screen warns that the target was not fully reached. Going past that would take a limiter, which changes the recording's sound and which this tool deliberately does not apply. A single gain also leaves the dynamics alone: the loud and quiet parts keep the same proportion.
MP3 adds one detail. The encoder can overshoot the input peak slightly when reconstructing the signal, which is why the ceiling is -1 dBFS and not 0. After extraction, the screen shows the real peak of what was written and how many samples hit the ceiling. The measured loudness can also drift a fraction of an LU between the computed value and the final file, because of lossy compression; in tests with a pure tone, the MP3 at 128 kbps landed about 0.5 LU below the prediction.

Gain computed in four situations

Quiet video with peak headroom

Input: -22 LUFS, pico -12 dBFS, alvo -14 LUFS
Expected output: +8 dB

The peak rises to -4 dBFS, under the -1 ceiling.

Quiet but with a high peak

Input: -24 LUFS, pico -3 dBFS, alvo -14 LUFS
Expected output: +2 dB (limitado pelo pico)

Stops at the -1 dBFS ceiling; loudness goes to -22 LUFS.

A video that is too loud

Input: -8 LUFS, pico -0,2 dBFS, alvo -14 LUFS
Expected output: -6 dB

Normalization also lowers.

Peak mode

Input: pico -10 dBFS, teto -1 dBFS
Expected output: +9 dB

Takes the peak to the ceiling, ignoring perceived loudness.

Full tool FAQ

How do I convert a video to MP3 without uploading the file?

Choose the video and click Extract. Reading, decoding and MP3 encoding all run in your browser, in a Worker, with no upload. The file only sits in your device's memory while it is processed.

Which MP3 bitrate should I pick?

Why does it say my video has no audio?

Can I extract just part of the video?

What does normalization do?

MP3, WAV or M4A: which format?

My video has 5.1 sound or 96 kHz audio. Now what?

Is there a size or length limit?

Which video formats are supported?

Does the MP3 get a title and artist?

Frequently asked questions

Which LUFS target should I choose?

-14 LUFS is the common streaming reference, -16 LUFS is often used for podcasts and -23 LUFS is the TV target under EBU R128. If you are not publishing anywhere, -16 is a good middle ground for listening on a phone.

Why did the tool say the target was not reached?

Because the section's peak would reach the -1 dBFS ceiling before the loudness reached the target, which means isolated peaks well above the average level. Going further would need a limiter. Trim out the part with the peak, or accept the level that fits.

Does normalizing change the sound quality?

A constant gain does not change timbre or dynamics; it just shifts everything up or down. Raising a quiet recording a lot also raises the background noise, and the tool caps the gain at 30 dB for safety.

Is the section's loudness measured separately from the rest?

Yes. The loudness is recomputed for the marked section, in 100 ms blocks, with the standard's gates (absolute -70 LUFS and relative -10 LU). Moving the handles updates the gain instantly.

Next steps

Continue in the tool

Open full tool

boost the volume of a video's audio mp3

Boost the volume of a video's audio while extracting it as MP3

Under More options, turn on Normalize the volume. The tool measures the section, shows the current loudness, the peak and the gain in dB, and applies a single gain while extracting.

Perceived loudness, peak and what "normalize" changes

Two numbers describe audio. The peak, in dBFS, is the highest sample: 0 dBFS is the digital maximum and going past it is distortion. Perceived loudness, in LUFS (ITU-R BS.1770), measures energy over time with a curve that mimics the ear and ignores silences; that is what tells you whether a video sounds quiet. A video can peak at -3 dBFS, looking "at the limit", and still sit at -24 LUFS, far below streaming's -14 LUFS reference. Raising volume without looking at the peak clips; looking only at the peak leaves the video quiet. LUFS normalization handles both.

The tool applies a constant gain: target minus measured loudness. The targets are -14 LUFS (streaming reference), -16 LUFS (podcast) and -23 LUFS (TV, EBU R128). The gain never takes the peak above -1 dBFS: if the target needs more, it stops at the ceiling and the screen warns that the target was not fully reached. Going past that would take a limiter, which changes the recording's sound and which this tool deliberately does not apply. A single gain also leaves the dynamics alone: the loud and quiet parts keep the same proportion.

MP3 adds one detail. The encoder can overshoot the input peak slightly when reconstructing the signal, which is why the ceiling is -1 dBFS and not 0. After extraction, the screen shows the real peak of what was written and how many samples hit the ceiling. The measured loudness can also drift a fraction of an LU between the computed value and the final file, because of lossy compression; in tests with a pure tone, the MP3 at 128 kbps landed about 0.5 LU below the prediction.

Gain computed in four situations

Quiet video with peak headroom

Input: -22 LUFS, pico -12 dBFS, alvo -14 LUFS
Expected output: +8 dB

The peak rises to -4 dBFS, under the -1 ceiling.

Quiet but with a high peak

Input: -24 LUFS, pico -3 dBFS, alvo -14 LUFS
Expected output: +2 dB (limitado pelo pico)

Stops at the -1 dBFS ceiling; loudness goes to -22 LUFS.

A video that is too loud

Input: -8 LUFS, pico -0,2 dBFS, alvo -14 LUFS
Expected output: -6 dB

Normalization also lowers.

Peak mode

Input: pico -10 dBFS, teto -1 dBFS
Expected output: +9 dB

Takes the peak to the ceiling, ignoring perceived loudness.

Full tool FAQ

How do I convert a video to MP3 without uploading the file?

Choose the video and click Extract. Reading, decoding and MP3 encoding all run in your browser, in a Worker, with no upload. The file only sits in your device's memory while it is processed.

Which MP3 bitrate should I pick?

Why does it say my video has no audio?

Can I extract just part of the video?

What does normalization do?

MP3, WAV or M4A: which format?

My video has 5.1 sound or 96 kHz audio. Now what?

Is there a size or length limit?

Which video formats are supported?

Does the MP3 get a title and artist?

Frequently asked questions

Which LUFS target should I choose?

Why did the tool say the target was not reached?

Does normalizing change the sound quality?

Is the section's loudness measured separately from the rest?

Yes. The loudness is recomputed for the marked section, in 100 ms blocks, with the standard's gates (absolute -70 LUFS and relative -10 LU). Moving the handles updates the gain instantly.